This article was originally published on the Journal - Law Society of Scotland website, on the 6th October 2026. Written by: Rachael Jane Ruth and Stephanie Barratt, Brodies LLP (with input from Martin Sloan, Partner, and James MacGregor, Ethical eDiscovery)
AI has moved firmly into the legal mainstream. It is reshaping how law firms search, draft, review and analyse information, whilst also changing how individuals pursue claims and exercise legal rights. Data Subject Access Requests (DSARs) are a clear example of that shift: AI is not only influencing how requests are made, but also how organisations may choose to respond.
Individuals have the right to ask an organisation to provide the personal data it holds about them. This right is set out in the UK General Data Protection Regulation (GDPR) and organisations typically have a short timeframe to respond. Many find the challenge lies in locating, even before reviewing and disclosing, relevant personal data. Modern businesses hold information across numerous systems, including email servers, document management platforms, collaboration tools, messaging applications and cloud storage environments.
For law firms responding to DSARs, AI cuts both ways. Used well, it can help identify personal data, review that information and organise information more efficiently. On the other side, AI is changing the data-related requests organisations receive. Individuals are using AI to generate broad and numerous DSARs which can be more time consuming and complex to deal with.
AI Generated DSARs
What previously may have required time, careful consideration and a broad legal understanding can now be generated in seconds using widely available generative AI tools which draw on public guidance and data protection legislation. ChatGPT, increasingly used to generate letters of claim or expressions of dissatisfaction, may default to include wording which the cautious organisation would read and treat as a subject access request.
The commercial implications are significant. AI use by individuals is likely to increase the volume and complexity of requests due to ease and accessibility. For law firms, more DSARs means more pressure to respond in a short timeframe. AI-generated DSARs are often broad, appear polished (at least on the face of it) and catch-all in nature. A requestor may only be seeking a few documents, but without careful prompting the AI output will require an organisation to produce a significant volume of documentation. This means that it requires searches across multiple systems, data sources and communication channels, increasing cost and review time.
AI-generated DSARs often contain extensive lists of demands, legal terminology and references to legislation. Practically speaking, however, it is important not to let AI “set the agenda”. Law firms have an obligation to carry out reasonable and proportionate searches to find the requestor’s personal data. While the requestor may provide helpful information that may assist in establishing the parameters of a search (for example, time periods or personal data relating to a particular issue or event), it is for the law firm to determine what searches are reasonable and proportionate, having regard to the nature of the information that it holds.
Law firms should respond to the substance of the request and apply the same legal and operational framework that would apply to any other DSAR. It is also important to define the scope of the DSAR early, break the request down into its constituent parts and identify precisely what personal data is being sought. If the request is unclear or so broad that clarification is reasonably required to provide an effective response, consider engaging with the requester in productive discussion at an early stage.
Responding to DSARs using AI and technology
AI and e-discovery platforms may provide the solution organisations need to process large volumes of data to respond to a specific request. Used well, this can be a win-win: individuals receive a quicker response; and firms minimise the time spent locating and identifying qualifying documents.
Ethical E-Discovery is a legal technology consultancy providing expert support to law firms. It is often instructed to advise on the best tool for the job in the context of litigation, arbitration and a variety of complex and high-profile disputes where efficient document review and analysis is essential. Founder of Ethical E-Discovery, James MacGregor’s experience is that e-discovery professionals are well-practiced in the use of AI, as machine learning has played a role in document review within e-discovery software for more than a decade. Predictive Coding, an early form of AI, is a practice whereby the software predicts whether a document would be coded as “relevant” or “not relevant” based on how previous documents had been coded. Once the system predicts it has found all the “relevant” documents (within an acceptable margin of error), reviewers can halt the process, removing the need to review large numbers of irrelevant documents. That process (which is still followed under certain circumstances) requires an experienced lawyer conducting the review of a statistically significant sample of the documents to train the algorithm, and so whilst it may save significant time, it isn’t always the most cost-effective option when responding to a DSAR.
In the generative AI era, it is now possible to conduct a complete first-level review of a DSAR without a lawyer reviewing a single document. Instead, a prompt can be written to define relevance; that prompt can then be run across all documents in the platform. The AI will classify the documents according to the prompt’s instructions and provide a justification for this classification, meaning that AI can be used to respond to an AI-generated DSAR.
A word of caution, however: firms’ data protection officers and those handling subject access requests must ensure they are familiar with the categories of data exempt from disclosure codified in the GDPR and Information Commissioner’s Office guidance.
Human review is essential when using AI to respond to a DSAR - documents may be incorrectly categorised, relevant personal data may be overlooked, may contain personal data relating to third parties (disclosure of which could breach the law firm’s obligations to those third parties),or exempt material may be inadvertently disclosed (including information that is subject to privilege or client confidentiality obligations). Any such errors may expose organisations to complaints, regulatory scrutiny or litigation. This is of particular importance when responding to a DSAR in the context of a complaint or claim, where careless disclosure of exempt material could prejudice future handling.
As organisations deploy AI tools more widely, questions arise as to whether AI systems themselves must be searched when responding to a DSAR. In many cases, the answer could be “yes”, where prompts, outputs, transcripts or other records generated by the preferred system may contain personal data relating to the requester. Data controllers and those handling DSARs must ensure a comprehensive search of all systems to ensure full compliance with the request.
Lawyers will be familiar with recent cases demonstrating that AI systems can produce inaccurate or fabricated outputs. When handling personal data, inaccuracies in a response could undermine compliance and potentially create additional risk. Using AI and legal technology might be a game-changer for your organisation in ensuring prompt and full responses to DSARs but ensure you keep the “human in the loop” to maintain compliance and keep service standards high.